INDEX

Explanations

mentions of the word "Pri" with varying intensities

oai_token-act-pair · gpt-3.5-turbo

mentions of prestigious awards or prizes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enegger

-0.88

icago

-0.83

lihood

-0.69

 Beckham

-0.69

ickr

-0.68

 bracelet

-0.67

schild

-0.66

dash

-0.65

00200000

-0.65

ç¥ŀ

-0.64

POSITIVE LOGITS

Pri

1.24

Pri

1.08

ests

0.96

esses

0.94

zes

0.91

pri

0.90

ety

0.88

ennes

0.88

vy

0.85

oshenko

0.84

Activations Density 0.006%

mentions of the word "Pri" with varying intensities

mentions of prestigious awards or prizes

No Comments

No Known Activations